U.S. patent number 7,051,269 [Application Number 09/958,084] was granted by the patent office on 2006-05-23 for method for channel coding.
This patent grant is currently assigned to Siemens Aktiengesellschaft. Invention is credited to Thomas Hindelang, Thomas Stockhammer, Wen Xu.
United States Patent |
7,051,269 |
Hindelang , et al. |
May 23, 2006 |
Method for channel coding
Abstract
A method for channel coding in a digital telecommunication
system, the method having error protection coding and interleaving,
a coherent puncturing and interleaving rule determined in a common
calculation step being applied for puncturing of a predetermined
error protection coding master code and the interleaving, the rule
having been determined under the criterion of optimizing the
distance properties after interleaving.
Inventors: |
Hindelang; Thomas (Munich,
DE), Stockhammer; Thomas (Bergen, DE), Xu;
Wen (Unterhaching, DE) |
Assignee: |
Siemens Aktiengesellschaft
(Munich, DE)
|
Family
ID: |
7903785 |
Appl.
No.: |
09/958,084 |
Filed: |
April 5, 2000 |
PCT
Filed: |
April 05, 2000 |
PCT No.: |
PCT/DE00/01055 |
371(c)(1),(2),(4) Date: |
October 04, 2001 |
PCT
Pub. No.: |
WO00/62425 |
PCT
Pub. Date: |
October 19, 2000 |
Foreign Application Priority Data
|
|
|
|
|
Apr 7, 1999 [DE] |
|
|
199 15 687 |
|
Current U.S.
Class: |
714/790 |
Current CPC
Class: |
H03M
13/23 (20130101); H03M 13/6362 (20130101); H03M
13/2792 (20130101) |
Current International
Class: |
H03M
13/23 (20060101); H03M 13/27 (20060101) |
Field of
Search: |
;714/790 |
References Cited
[Referenced By]
U.S. Patent Documents
Other References
XP-002133421--"Turbo" Decoding with Unequal Error Protection
applied to GSM speech coding--Burkert et al., pp. 2044-2048. cited
by other.
|
Primary Examiner: Baker; Stephen M.
Attorney, Agent or Firm: Bell Boyd & Lloyd LLC
Claims
What is claimed is:
1. A method for channel coding of speech signals for speech
transmission in a mobile radio system, the method including error
protection coding and interleaving, the method comprising:
determining a coherent puncturing and interleaving rule with a
common calculation, the rule being determined under a criterion of
optimizing distance properties after interleaving; and applying the
rule for puncturing of a predetermined error protection coding
mother code and for the interleaving, wherein the coherent
puncturing and interleaving rule is applied as one of a set of
subcodes and puncturing tables.
2. A method for channel coding as claimed in claim 1, wherein the
method is applied to a speech transmission method with blockwise
coding and with division of frames into a plurality of
timeslots.
3. A method for channel coding of speech signals for speech
transmission in a mobile radio system, the method including error
protection coding and interleaving, the method comprising:
determining a coherent puncturing and interleaving rule with a
common calculation, the rule being determined under a criterion of
optimizing distance properties after interleaving; and applying the
rule for puncturing of a predetermined error protection coding
mother code and for the interleaving, wherein the coherent
puncturing and interleaving rule is applied as one of a combined
puncturing and interleaving matrix and a table.
4. A method for channel coding of speech signals for speech
transmission in a mobile radio system, the method including error
protection coding and interleaving, the method comprising:
determining a coherent puncturing and interleaving rule with a
common calculation, the rule being determined under a criterion of
optimizing distance properties after interleaving; and applying the
rule for puncturing of a predetermined error protection coding
mother code and for the interleaving; determining a plurality of
coherent puncturation and interleaving rules in a plurality of
common calculation steps using a plurality of predefined
algorithms; performing a weighting step for determination of at
least one of a free distance and a mean distance spectrum, and
using the spectrum to determine associated distance properties; and
applying the coherent puncturation and interleaving rule in channel
coding which supplies at least one of a most favorable free
distance and a most favorable distance spectrum.
5. A method for channel coding of speech signals for speech
transmission in a mobile radio system, the method including error
protection coding and interleaving, the method comprising:
determining a coherent puncturing and interleaving rule with a
common calculation, the rule being determined under a criterion of
optimizing distance properties after interleaving; and applying the
rule for puncturing of a predetermined error protection coding
mother code and for the interleaving, wherein the method is applied
to a transmission channel with an approximately constant fading
amplitude within a time slot, and a transmission method in which
each new time slot is assigned another channel.
6. A method for channel coding of speech signals for speech
transmission in a mobile radio system, the method including error
protection coding and interleaving, the method comprising:
determining a coherent puncturing and interleaving rule with a
common calculation, the rule being determined under a criterion of
optimizing distance properties after interleaving; and applying the
rule for puncturing of a predetermined error protection coding
mother code and for the interleaving, wherein the method is applied
to a transmission channel with main error event extinction within a
timeslot.
Description
BACKGROUND OF THE INVENTION
Modem digital telecommunication systems provide source coding of
signals (for example, languages) with the aid of which compression
of the signal is achieved such that a substantially lower bit rate
than that of the (primary) digital signal suffices for
transmission. Thus, in accordance with the system standard of the
generally known GSM mobile radio system, the speech signal is
sampled at the transmitter end at a rate of 8000 samples per
second, the samples being represented with a resolution of 13 bits.
This corresponds to a bit rate of 104 kbit/s per speech signal. The
source or speech encoder compresses this speech signal to a
source-coded speech signal with blocks of lengths 260 bits and a
bit rate of 13 kbit/s. A compression of the speech signals by the
factor 8, therefore, takes place.
The physical conditions in the case of wireless telecommunication,
specifically in the frequency bands, available for mobile radio
systems, under terrestrial conditions in the case of which there is
multiple scattering and reflection on natural obstacles, lead to
high and relatively strong fluctuating propagation losses and
fading produced by multipath propagation (fast fading). In
individual time sections (timeslots) of the transmission process,
the transmission can thereby be strongly disturbed or even
completely interrupted, while other timeslots are, by contrast,
scarcely disturbed.
The useful datastream therefore, includes phases which either have
a higher or lower bit error rate; that is, the errors occur in
bursts, in particular.
Because of these circumstances, the transmission of the speech
signal, which is highly compressed and reduced in redundancy by the
source coding, will not be possible directly with acceptable
quality. The bit error rate to be expected (of the order of
magnitude of 10.sup.-3 to 10.sup.-1) is, therefore, to be reduced
to acceptable values (of the order of magnitude of 10.sup.-3 to
10.sup.-6) by suitable error correction methods. This is the task
of channel coding, which basically adds a (defined) redundancy once
again to the source coded signals, which then permits the detection
and correction of transmission errors on the transmission link (air
interface).
In the method of the generic type, channel coding includes error
protection coding (convolutional coding) and interleaving (also
denoted as scrambling). Subsequently, the convolutionally coded and
interleaved blocks are encrypted, mapped onto data bursts,
modulated onto the carrier frequency, and transmitted.
It must be kept in mind for convolutionally coding that the source
coded bits are not of equal relevance for the speech quality after
decoding. Errors in some bits lead to substantial impairments of
comprehensive validity, whereas errors and other bits are scarcely
perceptible. The source coded bits are therefore split into pulses
or groups, where each pulse or group is provided with a different
error protection. Thus, in the case of GSM full-rate encodec
(encoder/decoder for full rate transmission) they are the
protection classes 1a, 1b, and 2.
A conventional method for implementing this different protection is
what is termed "puncturation" or puncturing of the code following
upon the error protection coding (convolutional code). Simply, the
puncturation eliminates one or more positions from the output bit
stream of the convolutional encoder in accordance with a prescribed
scheme (a puncturation table). A puncturation table consists of
elements 0 and 1, and is periodically processed, the bit
corresponding to a 0 not being sent in the output bit stream, and
the bit corresponding to a 1 being transmitted. The coded sequence
is consequently shortened, and the error protection effect is
weakened. Punctured codes have the advantage, however, of the
implementability of various coding rates, codes with a higher
coding rate being developed by periodic puncturation starting from
a mother code of rate 1/n.
The fundamental sensitivity of the convolutional coding and
decoding methods to errors occurring in bursts, such as those which
constitute the main problem in the case of mobile radio
connections, is still further sharpened by the puncturation such
that punctured convolutional codes are applied virtually only in
conjunction with subsequent nesting or interleaving. An
approximately constant fading amplitude occurs within a timeslot in
the case of typical mobile radio channels with the profiles TU
(Typical Urban) or HT (Hilly Terrain) in the GSM system or in
similarly profiled channels. If another channel is used for each
new timeslot (ideal frequency hopping), the fading amplitude can be
assumed to be statistically independent between two timeslots.
Consideration as interleaving methods is given for this purpose to
block interleaving or random interleaving, for example, which are
known as such to the person skilled in the art.
The determination of effectively punctured convolutional codes
likewise constitutes an important design task for a digital
telecommunication transmission system of the type outlined above
such as the selection and concrete configuration of the
interleaving method to be applied in accordance with the error
protection coding. Systematic instructive methods are not known for
this purpose, and so it is necessary in order to find effectively
punctured convolutional codes to conduct computer-aided testing of
various codes and puncturings and compare them with one another
with the aid of specific criteria which permit a statement on the
likely transmission quality. It is typical, in this case, to
presuppose that the interleaver is ideal ("infinitely deep").
It is an object of the present invention, therefore, to specify a
method for channel coding which is improved with regard to the
achievable transmission quality.
SUMMARY OF THE INVENTION
The present invention is thus directed toward applying, in channel
coding, a coherent puncturation and interleaving rule, determined
in a common calculation step, for punctuating an error protection
coding mother code and for interleaving. A determination of this
coherent puncturation and interleaving rule proceeds from the
assumption of an ideal interleaver, and the configuration of the
interleaving step is put from the very start into an indissoluble
relationship with the finding of an advantageous puncturation rule
(puncturation matrix or puncturation table).
The puncturation and interleaving rule is summarized, in particular
in a predetermined set of subcodes and/or puncturation tables, or
else in a combined puncturation and interleaving matrix or table.
These are found in a computer-aided fashion observing the criterion
of optimization of the distance properties, specifically the
minimum free distance and/or the mean distance spectrum. These
terms are to be understood as follows: The minimum free distance
d.sub.f of a code is the minimum Hamming distance which occurs
between divergent paths. The mean distance spectrum {a.sub.d} is
given by {a.sub.d}=1/P.SIGMA..sup..infin.d=d.sub.fa.sub.d, a.sub.d
being the number of all paths; which, starting from P consecutive
nodes, leave the zero path, meet each other again and, in so doing,
build up the distance d and P being the number of columns in the
puncturation matrix P, and thus specifying the period length of the
puncturation. The mean distance spectrum {C.sub.d} is given by
{C.sub.d}=1/P.SIGMA..sup..infin.d=d.sub.fC.sub.d, the information
weight C.sub.d being the number of bits with the value 1 along all
paths which, starting from P consecutive nodes, leave the zero
path, meet again and in the process build up the distance d.
The bit error probability can be estimated for an AWGN (Additive
White Gaussian Noise) channel or a fully-interleaved Rayleigh
channel from the distance spectra. The optimum non-systematic
non-recursive convolutional codes and systematic recursive
convolutional codes for transmitting without puncturation at a rate
of 1/2 and memory 2 via the above named channels are known.
Thus, in practice, a number of puncturation and interleaving
matrices or tables are determined in a number of calculation steps
using a computer in accordance with a number of algorithms, their
distance properties are respectively determined in a weighting step
in accordance with the partial criteria, and that puncturation and
interleaving matrix or table is implemented in the channel coder
which shows the most favorable result with regard to the bit error
rate to be expected. It is particularly advantageous to apply the
proposed method for a transmission channel with the profile
mentioned further above, that is to say with an approximately
constant fading amplitude within a timeslot, and to a transmission
method in which each new timeslot is assigned another channel. The
advantages come to bear, in particular, in the case of application
to a speech transmission method with a high transmission rate. In
the case of this method, it is assumed that the failure of a
timeslot is an error event that occurs most frequently.
Investigations have also confirmed this.
Additional features and advantages of the present invention are
described in, and will be apparent from, the following Detailed
Description of the Invention and the Figures.
BRIEF DESCRIPTION OF THE FIGURES
FIG. 1 shows comparative diagrams of block error rates of a
half-rate channel for an interleaver determined by a conventional
method and of an optimized interleaver.
FIG. 2 shows comparative diagrams of bit error rates in the case of
a conventional interleaver and an optimized interleaver.
BRIEF DESCRIPTION OF THE INVENTION
A systematic, punctured code of rate R=11/16 is yielded by the
puncturation of a mother code of rate 1/2 and memory m=4 with the
aid of the generator polynomials .times. ##EQU00001##
The puncturation matrix, found via computer search, with the best
distance properties is as follows: ##EQU00002##
The period length P is 11 in this example. When selecting the
punctured codes, it must be kept in mind that the number Z of ones
in the puncturation matrix is an intger multiple of the number
timeslots in order to be able to assign each timeslot the same
number of bits. The interleaving assigns the bits of the matrix P
to the timeslots. In the example, the half-rate channel is assumed,
and the Z=16 bits are thereby split up into F=4 timeslots. The
following combined puncturation and interleaving matrix is obtained
if block interleaving (conventional independent interleaver) is
applied: ##EQU00003##
Elements of the matrix with the value 0 are punctured bits, and the
values 1.4 assign the individual bits to the corresponding
timeslots.
The failure of a block can now be regarded as a further
puncturation; that is, all the bits of this block are ignored. The
number of the lost bits is B=Z/F in the event of a block failure. A
new subcode with the rate: R.sub.u=P/Z-b, results, P being the
number of columns in the puncturation matrix P, and thus specifying
the period length of the puncturation.
The interleaver determines the configuration of the subcode by
assignment of the bits to the blocks of the channel. The selection
of the interleaver must be done such that the subcodes resulting
from block failure are optimum for the purpose of the described
distance criteria. A third search must be carried out after the
computer-aided search of the mother code of rate 1/2 and the
punctured code of rate 11/16. F good puncturation matrices must be
sought for the subcodes, their elements always being zero when the
puncturation matrix of the mother code contains a zero at this
point, and which may contain a zero at other positions only when no
other matrix of a subcode is zero at this point. Each bit must be
assigned to exactly one timeslot.
The following optimum subcodes of rate R.sub.u=11/12 are yielded
the assignment of the code of rate 11/16 to 4 timeslots.
##EQU00004## ##EQU00004.2## ##EQU00004.3## ##EQU00004.4##
These submatrices can be summarized to form the combined
puncturation and interleaver matrix: ##EQU00005##
This matrix differs from that of the block interleaver from (3.3).
The distance properties are better than those of the block
interleaver. The distance spectra of the individual puncturation
matrices of the subcodes are added in order to determine the
optimum code.
TABLE-US-00001 TABLE 3.1 Free distance and distance spectrum of the
punctured code of rate 11/16. d.sub.f = 4 a.sub.4 = 6 c.sub.4 = 18
a.sub.5 = 34 c.sub.5 = 132
TABLE-US-00002 TABLE 3.2 Free distance and distance spectrum of the
subcodes of rate 11/12. Code with d.sub.f = 2 a.sub.2 = 6 c.sub.2 =
21 a.sub.3 = 112 c.sub.3 = 556 P.sub.u1: Code with d.sub.f = 2
a.sub.2 = 4 c.sub.2 = 15 a.sub.3 = 78 c.sub.3 = 441 P.sub.u2: Code
with d.sub.f = 2 a.sub.2 = 4 c.sub.2 = 16 a.sub.3 = 106 c.sub.3 =
661 P.sub.u3: Code with d.sub.f = 2 a.sub.2 = 6 c.sub.2 = 26
a.sub.3 = 132 c.sub.3 = 807 P.sub.u4: Sum 20 78 428 2465
By comparison therewith, in the case of the block interleaver there
exists a subcode with a minimum distance d.sub.f=1. This leads to a
higher error rate.
FIG. 1 compares the block error rates of the half-rate channel for
the interleaver determined by the conventional method (dashed line,
denoted by "regular") and of the optimized interleaver, described
by the combined puncturation and interleaver matrix according to
equation (3.5), (continuous line, denoted by "optimum"). Even in
this simple example, a gain of 0.3 dB results, which can be
implemented without additional outlay at the decoder end.
FIG. 2 shows a comparison of the bit error rates in the case of a
conventional interleaver (block interleaver--see in FIG. 2 the
curves denoted by "old code") and an optimized interleaver (see in
FIG. 2 the curves denoted by "new code"). FIG. 2 relates to a
half-rate channel for the rate 8.1 kbit/s ("mode 3 H"), and the
code rates specified in the box, and shows the bit error rate
plotted against the bit number.
In this case, the following rates were used to code in order to
achieve an unequal error protection:
TABLE-US-00003 TABLE 1 Rates of the mode 3H with block and
optimized interleaver Block interleaver Block Bit No. 0.1 2.5 6.23
24.47 48.95 96.163 interleaver Number 2 4 18 24 48 68 Rate 1/3 2/5
1/2 8/14 8/11 1 Optimum Bit No. 0.1 2.5 6.22 23.42 43.97 98.163
interleaver Number 2 4 17 20 55 66 Rate 1/3 2/5 1/2 10/16 11/16
1
Investigations by the present inventors have shown that the gain
which can be achieved with the aid of the proposed method is
greater the higher the rate of the code. Again, increasing the
number Z of bits which are combined leads to an improvement in the
distance properties, and thus to a reduction in the bit error rate
to be expected.
Although the present invention has been described with reference to
specific embodiments, those of skill in the art will recognize that
changes may be made thereto without departing from the spirit and
scope of the invention as set forth in the hereafter appended
claims.
* * * * *